One of the most significant manufacturing issues is how to monitor and diagnose the state of machines from various sensor data. Detecting machine state changes is very important because it can prevent machine breakdown or product quality deterioration. This work deals with a change point detection problem of semiconductor manufacturing equipment, especially for a deposition process, with real data. In a deposition machine, the pressures of two pumping lines, each of which absorbs a different gas (i.e., Zr and ozone), should be well-balanced so that the two gases are not mixed in a line. Otherwise, particles can be accumulated on the surface of wafers, which significantly affects the product quality. When one gas is pumped into another gas line, the balance of pumping lines is considered to be broken. We propose a l(1) trend filtering-based change point detection method to identify such a pumping balance break in the deposition machine. This method is suitable for data having increasing or decreasing trends, such as pressure and temperature, and can be applied for the real-time detection. The proposed method shows an adequate true detection rate while effectively reducing the number of false alarms compared to other methods. We further develop a prediction model for estimating the pressure in the machine to improve the performance of the change point detection algorithm.